Bad beliefs: automaticity, arationality, and intervention

ABSTRACT

Levy (2021) argues that bad beliefs predominately stem from automatic (albeit rational) updating in response to testimonial evidence. To counteract such beliefs, then, we should focus on ridding our epistemic environments of misleading testimony. This paper responds as follows. First, I argue that the suite of automatic processes related to bad beliefs extends well beyond the deference-based processes that Levy identifies. Second, I push back against Levy’s claim that bad beliefs stem from wholly rational processes, suggesting that, in many cases, such processes are better characterized as arational. Finally, I note that Levy is too quick to dismiss the role that individuals can play in cleaning up their own epistemic environments, and I suggest one route through which this is possible.

Keywords:

• Advocacy Effect
• arationality
• bad Beliefs
• irrationality
• rationality
• truth repetition effect

Acknowledgments

Thanks to Dan Williams for useful feedback. This research was funded by a Fonds Wetenschappelijk Onderzoek (FWO) Postdoctoral Fellowship (1267022N).

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. At various points, Levy argues that social referencing—“looking to others within our social group for cues for what to believe” (p. 48)—also plays an important role. It seems to me, however, that social referencing (defined in this way) merely precedes deference; it allows us to identify the appropriate testimonial evidence to update off. Consequently, I will focus on the more essential stage of (deferential) updating.

2. The smooth and automatic nature of deference isn’t only endorsed by Levy, it is essential to his argument. Levy claims that deference facilitates the outsourcing of knowledge generation. Outsourcing functions to reduce the complexity of individual cognizing. However, carefully reasoning about matters in a way that heavily weights testimonial evidence doesn’t facilitate outsourcing because it doesn’t reduce the amount of cognitive work involved.

3. Another possibility is that, in preparing to advocate for or against the client, participants generated arguments to prepare, and thinking of such arguments acted as a form of self-persuasion. However, the act of defending or prosecuting Harris did not involve arguments whatsoever. Instead, participants simply prepared to play a game called “attorney at law” which involved quickly pressing the space bar when the word GUILTY (for defenders) or INNOCENT (for prosecutors) appeared on the screen. Given the nature of this game, it is unlikely participants would have generated arguments for their position.

4. Some suggest that belief updating to testimonial evidence is irrational in this sense, though such results are open to debate (Tappin & Gadsby, 2019).

Additional information

Funding

The work was supported by the Research Foundation Flanders [1267022N].